Range of Risk Factor Levels, Risk Control, and Temporal Trends for Nephropathy and End-stage Kidney Disease in Patients With Type 1 and Type 2 Diabetes

Diabetic nephropathy is one of the most serious complications of diabetes and the leading cause of end-stage kidney disease (ESKD) (1–3). Several modifiable and nonmodifiable risk factors have been associated with increased risk of diabetic nephropathy, such as duration of diabetes, hypertension, hyperglycemia, and obesity (4–6). Previous research has demonstrated the importance of managing risk factors, as better glycemic control, lipid levels, and normalized blood pressure have been shown to reduce the risk of onset and progression of diabetic nephropathy (7–10).

Over the past decades, there have been numerous advances in the management of diabetes and related risk factors, such as the introduction of sodium–glucose cotransporter 2 inhibitors and glucagon-like peptide 1 analogs in type 2 diabetes and increased use of modern insulin-delivering technologies and sensors in type 1 diabetes, as well as improved control of blood pressure and lipids (11,12). The recent improvements in diabetes care have ultimately resulted in declining rates of mortality and several diabetes-related complications, such as acute myocardial infarction, stroke, and amputations (13,14). However, data on long-term trends for diabetic nephropathy and ESKD are limited. Despite the clinical improvements, a substantial excess risk for ESKD remains for patients with diabetes compared with individuals without diabetes (14–16). Few studies have included investigation of whether excess risk of kidney impairment in patients with diabetes is affected by differences in risk factor control or assessment of the optimal levels of traditional and modifiable cardiometabolic risk factors in the prevention of kidney complications. Moreover, few studies have explored the long-term trends in kidney complications with use of data from large and well-defined cohorts of patients with type 1 and type 2 diabetes and matched control subjects with two decades of follow-up.

In this study we sought to investigate long-term trends in incidence rates of diabetic nephropathy and ESKD over the past two decades in people with type 1 and type 2 diabetes from the Swedish National Diabetes Register (NDR), compared with control subjects randomly selected from the general population. Additionally, excess risk of ESKD in diabetes, the relative risk according to degrees of multifactorial risk factor control, compared with matched control subjects, and optimal levels for selected cardiometabolic risk factors were assessed.

This nationwide observational study was approved by the Ethical Review Authority (Lund, Sweden; 2020-04796). All patients provided informed consent prior to entry into the registry.

In this study we use data from the NDR, which has previously been described. Distinction between type 1 and type 2 diabetes is based on the epidemiological definitions and physicians’ clinical assessment (17,18). Simultaneously with the first registration in NDR, five control person subjects without diabetes, matched for age, sex, and county, were randomly selected for every study participant with diabetes by Statistics Sweden. Patients with at least one observation in the NDR between 1 January 2001 and 31 December 2019 were included in the study. Patients with diabetes who had diabetic nephropathy or ESKD at baseline (see Supplementary Table 2) were excluded from the study along with their selected control subjects. In addition, the control subjects who fulfilled any of the exclusion criteria were excluded separately. Study participants with diabetes and valid imputation data were included in the final cohort, along with their matched cohorts.

We assess two kidney outcomes: a composite outcome of kidney impairment (consisting of acute and chronic kidney failure, dialysis, and kidney transplantation among others), which we refer to as ESKD, and diabetic nephropathy. The outcome data were based on the main diagnosis and up to six secondary diagnosis codes from the Swedish inpatient and outpatient registry, using diagnosis codes of ICD-10. The specific codes included in each outcome measure are listed in Supplementary Table 2. Patients were followed until 31 December 2019 or until an event or death occurred.

The study interval from 2001 to 2019 was divided into 2-year periods, except for the last time period of the remaining 3 years. We adjusted incidence rates using direct standardization to the sex and age distribution of the first time period. We constructed the following five age categories for the study cohort: <45, 45–54, 55–64, 65–74, and >75 years of age. Incident rates were calculated as a ratio between the number of events that occurred during each period (numerator) and the number of individuals at risk during the same period (denominator). The numbers of events during each time period, as well as crude and standardized incidence rates as the number of events per 100,000 person-years, are presented in Supplementary Tables 4–7. Further, we also studied long-term trends in individuals with and without cardiovascular comorbidities at baseline.

The association between selected metabolic risk factors and the incidence of both outcomes was studied with Cox regression analyses, separately for participants with type 1 and type 2 diabetes. Data on metabolic risk markers were not available for the control subjects, who were consequently excluded from these analyses. Risk factors incorporated in this analysis included glycated hemoglobin (HbA_1c) (mmol/mol), BMI (weight in kilograms divided by square of height in meters), systolic blood pressure (SBP) (mmHg), diastolic blood pressure (mmHg), LDL cholesterol (LDL-C) (mmol/L), and HDL cholesterol (HDL-C) (mmol/L), and triglycerides (TRL) (mmol/L). We used 52 mmol/mol (6.9%) for HbA_1c, 130 mmHg for SBP, 1.6 mmol/L for HDL-C, 1.8 mmol/L for TRL, and 25 kg/m² for BMI. To enable the detection of nonlinear associations, we applied restricted cubic splines with three evenly spaced knots to the Cox regression models; additional knots for continuous predictors did not contribute particularly much to regression models. Further, the models were adjusted for the following covariates: age, sex, smoking, physical activity, ethnicity, marital status, income level, educational level, comorbidities, and pharmacological treatment of diabetes. These models also enabled us to estimate the values associated with the lowest risk for each outcome. An example of our Cox model, for estimating the optimal level of HbA_1c, is presented in Supplementary Material.

We constructed Cox models to investigate the association between the coexistence of multiple risk factors and risk of kidney failure and diabetic nephropathy. This was done by stratifying the patients with diabetes into groups according to the number of risk factors within target ranges (target levels recommended in clinical guidelines), at baseline. The risk factors were modeled as categorical variables (the risk factor within the target ranges: yes/no). The risk factors included in these models (with cutoff values in parentheses) were HbA_1c (≥7.0% or ≥53 mmol/mol), SBP (≥130 mmHg), smoking (being a current smoker at baseline), LDL-C (≥2.5 mmol/L or ≥97 mg/dL), and BMI (≥27.5 kg/m²). In the models assessing risk of kidney failure, we adjusted for duration of diabetes by assigning matched control subjects to a duration of zero years, while patients with diabetes had their duration of diabetes centralized around the grand mean. In the Cox models created for diabetic nephropathy, the patients with diabetes who had all risk factors within the target intervals were considered as a reference group, and the model was adjusted for duration of diabetes. The constructed Cox models were adjusted for age, sex, socioeconomic variables, baseline comorbidities, and treatment with antihypertensives, statins, or antithrombotic medications.

We created Cox models to estimate excess risk for ESKD in the entire diabetes cohorts, compared with control subjects, whereas risk for various risk factors associated with diabetic nephropathy are presented for patients with diabetes. In addition, these models were adjusted for age, sex, socioeconomic variables and comorbidities.

Missing data (∼5–10%) were handled with use of Multiple Imputation by Chained Equations (MICE). The distributions and means were analyzed before and after imputation, without observation of any significant differences. Variables included in the imputation model are presented in Supplementary Table 3. Statistical significance was considered to be indicated by a P value <0.05. A comprehensive presentation of statistical methods and model construction can be found in Supplementary Material. We performed calculations in R, version 4.01 (R Foundation for Statistical Computing), and analyses in RStudio.

The data used in this study are available from the sources stated in the article on request to the data providers, with fulfillment of the legal and regulatory requirements and approval from the Swedish Ethical Review Authority.

In all, 34,723 individuals with type 1 diabetes, along with 160,558 control subjects, and 666,899 patients with type 2 diabetes with 2,577,579 control subjects were included in the study. Mean age was 34 years for type 1 and 65 years for type 2 diabetes (Supplementary Table 1). Cardiovascular comorbidities were roughly twice as frequent in participants with diabetes, and they were also more often treated with anticoagulants, antithrombotic medications, statins, and blood pressure–lowering medications. Median follow-up was 10.0 years for type 1 diabetes and 7.5 years for type 2 diabetes. For the entire case-control cohort, 146,566 cases of ESKD and 17,534 cases of diabetic nephropathy were recorded during the follow-up period.

We observed an increasing trend of ESKD in both subjects with type 2 diabetes and matched control subjects. The incidence rate of ESKD (per 100,000 person-years) was 116.1 for type 2 diabetes and 42.0 for control subjects in the final time period (Fig. 1A). A similar trend was also observed in the study cohort where individuals with cardiovascular comorbidities at baseline were excluded (Fig. 1B). For diabetic nephropathy, a decreasing trend in incidence rates was observed in type 2 diabetes (Fig. 1C). The incidence rate was 76.0 cases/100,000 person-years during the first period and 23.5 cases during the last period.

For patients with type 1 diabetes, no clear secular trend in the incidence rates of ESKD was detected (Fig. 1E). However, incidence rates for control subjects displayed an increasing trend, with 8.2 cases/100,000 person-years during the first period and 51.4 cases during the final period. Further, the incidence of ESKD was significantly higher among those with type 1 diabetes compared with control subjects. The incidence rate during the final period was 155.9/100,000 person-years for patients with type 1 diabetes and 51.4 for control subjects. For diabetic nephropathy, a decreasing trend in the incidence rates was observed in type 1 diabetes (Fig. 1G). The incidence rate in the first period was 139.4 cases/100,000 person-years and 61.2 cases during the last period.

Figure 2 shows the associations between a series of cardiometabolic risk factors and ESKD for patients with type 2 diabetes. Elevated HbA_1c and SBP, higher BMI, and older age, as well as low levels of HDL-C and LDL-C, were associated with a higher risk for kidney failure. In addition, TRL levels below target guideline levels were associated with considerably lower risk. Similar patterns were observed for risk factors and diabetic nephropathy, although all associations were slightly less significant due to wider CIs.

In patients with type 1 diabetes, the interpretation of the associations was less clear due to wider CIs. High levels of HbA_1c and SBP and a longer duration of diabetes were associated with higher risk of both kidney failure and diabetic nephropathy for patients with type 1 diabetes (Fig. 3A and B). TRL levels below target guideline levels were associated with significantly lower risk of kidney failure and diabetic nephropathy. Furthermore, low BMI was associated with higher risk of kidney failure and diabetic nephropathy. As in type 2 diabetes, low levels of HDL-C and LDL-C were also associated with higher risk of kidney outcomes in type 1 diabetes. The association between kidney failure and age showed a monotonic linear relationship with a slight increase with increasing age, whereas increasing age was associated with a reduced risk of diabetic nephropathy.

Figure 4 displays adjusted hazard ratios (HRs) for ESKD and diabetic nephropathy for six subgroups of patients with diabetes, classified according to their risk factor control status, in comparison with matched control subjects. In type 2 diabetes, the risk of kidney failure was incrementally higher for each additional risk factor not within the target ranges. Compared with control subjects, participants with type 2 diabetes and five risk factors outside the target intervals had an HR of 3.61 (95% CI 3.44–3.79), while optimal risk factor control (i.e., no risk factors outside the target range) was associated with an HR of 1.60 (1.49–1.71). In type 1 diabetes, a similar pattern of rising HRs with an increasing number of uncontrolled risk factors was observed: patients with five risk factors outside of target range HR 12.67 (9.34–17.20) and patients with all five risk factors within target range HR 6.10 (4.69–7.93) (Fig. 4B).

For diabetic nephropathy, HR for patients with type 2 diabetes and five risk factors was 1.68 (95% CI 1.55–1.83) and for patients with one risk factor 1.01 (0.94–1.09). The risk of diabetic nephropathy increased by each risk factor outside the target range. A pattern of increasing HRs with increasing number of risk factors was also observed for patients with type 1 diabetes, though the CIs were wider. HR for patients with five risk factors was 2.42 (1.64–3.59), whereas HR for patients with one risk factor was 1.00 (0.76–1.33).

With adjustment for age, sex, and socioeconomic variables, HR for ESKD was 2.32 (95% CI 2.30–2.35) for patients with type 2 diabetes (Fig. 5A) and 10.92 (10.15–11.75) for patients with type 1 diabetes (Fig. 5B) compared with controls. A lower risk for kidney failure was observed in women and study participants of Scandinavian ethnicity in both type 1 and type 2 diabetes cohorts. All models revealed that increased levels of education were associated with lower risk for outcomes, and cardiovascular disease at baseline was associated with increased risk for outcomes in both patient groups, particularly for heart failure.

The results of this nationwide observational study of individuals with type 1 and type 2 diabetes, as well as matched control subjects from the general population, demonstrates that diabetic nephropathy has decreased substantially over the past two decades. However, incidence rates of kidney failure displayed a nonlinear pattern, with an initial decrease in rates and thereafter a gradual increase, for both patients with diabetes and control subjects. Patients with type 1 diabetes had greater absolute rates for developing kidney complications compared with patients with type 2 diabetes. This study also shows that the risk of diabetic nephropathy and ESKD increases in a stepwise fashion with the accumulation of risk factors and that the risk of kidney complications in patients with diabetes may be modifiable to a certain degree through comprehensive risk factor management. Patients with diabetes and optimal risk factor control continued to have excess risk for ESKD, and worsening risk factor control was associated with a substantial risk increase, compared with control subjects, whereas varying degrees of risk factor control did not alter the risk of diabetic nephropathy significantly. Traditional and modifiable cardiometabolic risk factors such as HbA_1c, SBP, BMI, TRL, and HDL-C were all associated with considerable increase in risk of ESKD and diabetic nephropathy. Our analyses reveal that levels lower than contemporary guideline levels were associated with lower risk for kidney complications among both patients with type 1 diabetes and patients with type 2 diabetes.

The declining rates of diabetic nephropathy observed in the current study are likely related to clinical advancements over the past decades, such as improved glycemic control, widespread use of antihypertensive medications and lipid-lowering agents, and improved smoking cessation. These clinical improvements likely played a role in the initial decline observed for incident ESKD, with several possible explanations for the increasing rates observed during latter time periods. Advances in clinical management of cardiometabolic disease have resulted in increased life expectancy and gradual progression of albuminuria and risk for kidney impairment. Furthermore, improvements in kidney-replacement therapy have enabled treatment for older and multimorbid patients (19,20). Decreasing incidence rates of diabetes-related kidney-replacement therapy have been observed in studies since the early 2000s (19,21,22). This study included a composite outcome for ESKD, composed of several diagnoses, in addition to dialysis and kidney transplantation. Frequent testing of kidney function, improved diagnostic methods, and earlier detection of kidney impairment have likely also contributed to the observed increment.

The excess risk for kidney failure in patients with diabetes compared with that for individuals without diabetes is in line with the results of previous studies (15,16,22,23). However, we also observed that patients with diabetes were at higher risk for ESKD despite optimal risk factor control.

It is well documented that hyperglycemia, SBP, lipids, BMI, and smoking are causal factors for kidney damage. The role of these risk factors in the progression of kidney dysfunction, through vascular damage, endothelial dysfunction, inflammation, oxidative stress, and accumulation of advanced glycation end products, is also well established (24–26). Consistent with the literature, our study shows that HbA_1c, SBP, and BMI were associated with increased risk of diabetic nephropathy and ESKD for patients with diabetes. HbA_1c displayed a clear linear pattern with significant relative risk reduction for levels below therapeutic target values in patients with diabetes, whereas risk reduction for SBP levels below target values was more prominent for patients with type 1 diabetes. BMI showed a J-shaped association for kidney outcomes in type 1 diabetes, which is presumably due to reverse causation. The reduced risk of diabetic nephropathy with increasing age is most likely due to diagnostic bias.

Previous observational studies have shown a clear association for TRL and HDL-C and risk for kidney disease (18,27–32). LDL-C has been associated with a higher risk of albuminuria in type 1 diabetes but not in type 2 diabetes (18,29). Low levels of HDL-C were also associated with increased risk of kidney outcomes in the current study. TRL levels below contemporary guideline levels were associated with significantly lower risk of diabetic nephropathy and ESKD, which reflects findings of previous studies emphasizing the role of hypertriglyceridemia for kidney dysfunction (30–32). In this study, decreasing levels of LDL-C were associated with higher risk of kidney outcomes and high LDL-C was associated with slightly lower risk of kidney outcomes in type 2 diabetes. This rather contradictory relationship between LDL-C and risk of kidney outcomes is likely attributable to reverse causality.

Cardiometabolic data were not available for the control subjects, making it impossible to assess the associations between these risk factors and kidney impairment in the general population or to investigate the extent to which the association between diabetes and kidney failure is attributable to such risk factors. Using baseline values of risk factors in analyses may be considered a limitation because, for some patients, baseline values will not be representative of follow-up. On the other hand, using index values is advantageous from a clinical perspective. The results of this study are model dependent and could vary slightly with different approaches to data analyses. No distinction was made between patients with all or some risk factors within target ranges without any specific intervention and patients who were medically treated to attain target levels.

No adjustments were made for multiple testing. We also acknowledge that residual confounding is impossible to fully overcome. Also, diabetic nephropathy is intermittently diagnosed with combinatorial ICD codes, in both inpatient and outpatient clinics, such as in the case of multiple diabetes-related microvascular complications.

This study contributes information on a changing epidemiology of kidney complications in diabetes: the incidence of diabetic nephropathy has declined over the past two decades, whereas kidney impairment is increasing for people with and without diabetes. The risk of diabetic nephropathy and kidney impairment in patients with diabetes is likely modifiable through multifactorial risk factor control. Compared with those without diabetes, individuals with diabetes are at higher risk for kidney impairment despite optimal control for selected cardiometabolic risk factors.

Funding and Duality of Interest. This work was supported by grants from the Swedish state under an agreement between the Swedish government and the county councils concerning economic support of research and education of doctors (ALFGBG-966187), the Swedish Heart and Lung Foundation (2019-0532), and the Swedish Research Council (2019-02019). D.L.B. discloses the following relationships: advisory board, AngioWave, Bayer, Boehringer Ingelheim, Cardax, CellProthera, Cereno Scientific, Elsevier Practice Update Cardiology, Janssen, Level Ex, Medscape Cardiology, Merck, MyoKardia, NirvaMed, Novo Nordisk, PhaseBio, PLx Pharma, Regado Biosciences, and Stasys Medical Corporation; board of directors, AngioWave (stock options), Boston VA Research Institute, DRS.LINQ (stock options), Society of Cardiovascular Patient Care, and TobeSoft; Inaugural Chair, American Heart Association Quality Oversight Committee; data monitoring committees, Acesion Pharma, Assistance Publique–Hôpitaux de Paris, Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute, for the PORtopulmonary Hypertension Treatment wIth maCitentan - a randOmized Clinical Trial [PORTICO trial], funded by St. Jude Medical, now Abbott), Boston Scientific (chair, Pulmonary Embolism Thrombolysis [PEITHO] trial), Cleveland Clinic (including for the CENTERA THV System in Intermediate Risk Patients Who Have Symptomatic, Severe, Calcific, Aortic Stenosis Requiring Aortic Valve Replacement [ExCEED] trial, funded by Edwards LifeSciences), Contego Medical (chair, Protection Against Emboli During Carotid Artery Stenting Using a 3-in-1 Delivery System Comprised of a Post-dilation Balloon, Integrated Embolic Filter and a Novel Carotid Stent II [PERFORMANCE II]), Duke Clinical Research Institute, Mayo Clinic, Mount Sinai School of Medicine (for the Edoxaban Compared to Standard Care After Heart Valve Replacement [ENVISAGE trial], funded by Daiichi Sankyo, and for the ABILITY-DM trial, funded by Concept Medical), Novartis, Population Health Research Institute, and Rutgers University (for the National Institutes of Health–funded Myocardial Ischemia and Transfusion [MINT] trial); honoraria, American College of Cardiology (Senior Associate Editor, Clinical Trials and News, ACC.org, and chair, American College of Cardiology Accreditation Oversight Committee), Arnold & Porter Kaye Scholer LLP (work related to Sanofi/Bristol-Myers Squibb clopidogrel litigation), Baim Institute for Clinical Research (Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention [RE-DUAL PCI] clinical trial steering committee, funded by Boehringer Ingelheim, and ApoA-I Event reducinG in Ischemic Syndromes II [AEGIS-II] executive committee, funded by CSL Behring), Belvoir Publications (editor in chief, Harvard Heart Letter), Canadian Medical & Surgical Knowledge Translation Research Group (clinical trial steering committees), Cowen and Company, Duke Clinical Research Institute (clinical trial steering committees, including for A Trial Comparing Cardiovascular Safety of Degarelix Versus Leuprolide in Patients With Advanced Prostate Cancer and Cardiovascular Disease [PRONOUNCE], funded by Ferring Pharmaceuticals), HMP Global (editor in chief, Journal of Invasive Cardiology), Journal of the American College of Cardiology (guest editor, associate editor), K2P (co-chair, interdisciplinary curriculum), Level Ex, Medtelligence/ReachMD (continuing medical education [CME] steering committees), MJH Life Sciences, Oakstone CME (Course Director, Comprehensive Review of Interventional Cardiology), Piper Sandler, Population Health Research Institute (for the Cardiovascular Outcomes for People Using Anticoagulation Strategies (COMPASS) operations committee, publications committee, and steering committee, and U.S. national co-leader, funded by Bayer), Slack Publications (chief medical editor, Cardiology Today’s Intervention), Society of Cardiovascular Patient Care (secretary/treasurer), WebMD (CME steering committees), and Wiley (steering committee); other, Clinical Cardiology (deputy editor), NCDR ACTION Registry Steering Committee (chair), and Veterans Administration Clinical Assessment, Reporting and Tracking System for Cath Labs (VA CART) Research and Publications Committee (chair); research funding, Abbott, Acesion Pharma, Afimmune, Aker BioMarine, Amarin, Amgen, AstraZeneca, Bayer, Beren, Boehringer Ingelheim, Boston Scientific, Bristol-Myers Squibb, Cardax, CellProthera, Cereno Scientific, Chiesi, CSL Behring, Eisai, Ethicon, Faraday Pharmaceuticals, Ferring Pharmaceuticals, Forest Laboratories, Fractyl Health, Garmin, HLS Therapeutics, Idorsia, Ironwood, Ischemix, Janssen, Javelin, Lexicon, Lilly, Medtronic, Merck, Moderna, MyoKardia, NirvaMed, Novartis, Novo Nordisk, Owkin, Pfizer, PhaseBio, PLx Pharma, Recardio, Regeneron, Reid Hoffman Foundation, Roche, Sanofi, Medical Corporation, Synaptic, The Medicines Company, and 89bio; royalties, Elsevier (editor, Braunwald’s Heart Disease); site co-investigator, Abbott, BIOTRONIK, Boston Scientific, CSI, Endotronix, St. Jude Medical (now Abbott), Philips, and Svelte; trustee, American College of Cardiology; and unfunded research, FlowCo and Takeda. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. The study was designed by J.H and A.R. J.H. wrote the first draft of the manuscript. Statistical analyses were performed by J.H. and A.R. A.R. vouches for the data and analysis. All of the authors participated in data analysis and interpretation. All authors vouch for the accuracy and completeness of the data and analyses and made the decision to submit the manuscript for publication. All named authors meet the International Committee of Medical Journal Editors criteria for authorship for this article, take responsibility for the integrity of the work as a whole, and gave approval for this version to be published. All of the authors had full access to the complete data in the study. A.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.